trading to hedge: dynamic hedging - new york universitypeople.stern.nyu.edu/jhasbrou/teaching/post...

1

4/16/2015

F1 Results

News vs. no-news

With news visible, the median trading profits were about $130,000 (485 player-sessions)

With the news screen turned off, median trading profits were about $165,000 (283 player-sessions)

Trading strategies that worked.

Market vs. limit order strategies.

Copyright 2015, Joel Hasbrouck, All rights reserved 2

2

4/16/2015

Entering limit and orders to buy


With V=100 and O(offset)=1, a left-click in this region will submit a limit order to buy 100 shares priced at 23.92 + 0.01 =23.93.The click inserts a new limit buy order at the indicated position.


To buy with a market order, you have to lift offers on the other side of the book.

You have to right click on the offer side to generate a market buy order.

3

4/16/2015

Entering limit and market orders to sell


To insert a new limit sell order, point to a row and left-click

To submit a market sell order, point anywhere in the bid side of the book and right-click.

Trading to hedge: dynamic hedging

4

4/16/2015

The RIT H1 portfolio hedge case: some key features

The risk is market risk in a known portfolio.

The hedging security is a stock index futures contract.

The relation between the portfolio return and futures return is linear, but partially random.

Partially random: we need to estimate beta from a statistical model and a data sample.

Linear: once we implement the hedge, we don’t have to adjust it. The hedge is static.


The RIT H3 case: an option hedge

We are short a risky security (a call option).

We will try to hedge by going long the stock.

The value of the call option is an exact, known function of the stock price (the Black-Scholes equation).

We don’t need to estimate a statistical model.

The Black-Scholes value of a call is nonlinear.

When the stock price changes, we need to adjust our hedge.

The hedge is dynamic.


5

4/16/2015

Review

An American call gives the holder the right to buy the underlying stock (𝑆) at exercise/strike price X up to and including maturity date T.

An American put gives the holder the right to sell the underlying …

A European option can only be exercised at maturity.

An option comes into existence when it is traded.

The person who sells a call has written the call, and is shortthe call


Intrinsic value of a call with exercise price 𝑋.

If the current stock price is 𝑆, the gain from immediate exercise is𝐼𝑛𝑡𝑟𝑖𝑛𝑠𝑖𝑐 𝑣𝑎𝑙𝑢𝑒 = 𝑀𝑎𝑥 𝑆 − 𝑋, 0


0 20 40 60 80S

10

20

30

40

Call with X 40

Max S X ,0

6

4/16/2015

𝐶, the value of the call

The call value is computed from the Black-Scholes equation.

The vertical difference between the two lines is the time value of the call.

Example: If 𝑋 = 40, 𝑆 = 48 and 𝐶 = 15, then the intrinsic value is𝑀𝑎𝑥 48 − 40,0 = 8

The time value is 15 − 8 = 7.13

0 20 40 60 80S

10

20

30

40

Call with X 40

Call Value

Max S X ,0

Call valuation: the Black-Scholes assumptions

The stock (and risk-free bonds) may be bought and sold at any time without cost. This allows us to construct and maintain perfectly hedged (risk-free)

portfolios of stocks, bonds and calls.

In fact, real-world options traders use approximate (low risk) hedges.

The stock price moves as the accumulation of small (infinitesimal) random changes that have constant volatility. There are no sudden announcements or surprises. In fact, the possibility of “jumps” can lead to discrepancies between

Black-Scholes valuations and actual market prices.


7

4/16/2015

Black-Scholes: the inputs

S the current stock price ($ per share) X the exercise price of the call option ($ per share) 𝑇 time to maturity (years) 𝑟 the risk-free interest rate (annual) A 4% rate would be entered as 𝑟 = 0.04

𝜎 the volatility of the underlying. The standard deviation of the stock’s annual return. Example: “Over time the S&P index average annual return is

about 10%, with a standard deviation of about 20%” This would be entered as 𝜎 = 0.20


The Black-Scholes equation for C, the value of a call

𝐶 = 𝑆𝐶𝑢𝑟𝑟𝑒𝑛𝑡𝑠𝑡𝑜𝑐𝑘𝑝𝑟𝑖𝑐𝑒

× 𝑁 𝑑1 − 𝑋𝐸𝑥𝑒𝑟𝑐𝑖𝑠𝑒𝑝𝑟𝑖𝑐𝑒

× 𝑒−𝑟×𝑇

𝑃𝑟𝑒𝑠𝑒𝑛𝑡𝑣𝑎𝑙𝑢𝑒𝑓𝑎𝑐𝑡𝑜𝑟

× 𝑁(𝑑2)

r is the (risk-free) interest rate for borrowing and lending.

T is the time remaining to maturity.

𝑑1, 𝑑2, and 𝑁(∙) are given on the next slide.

This variant of the equation is correct for a European call on a non-dividend paying stock.


8

4/16/2015

𝑑1 =𝑙𝑛𝑆

𝑋+ 𝑟+𝜎2

2𝑇

𝜎 𝑇

𝑑2 = 𝑑1 − 𝜎 𝑇

𝑁(𝑑) is the cumulative distribution for the standard normal density evaluated at 𝑑.

𝑁 𝑑 is given by the Excel function NORM.S.DIST(d,TRUE)

17

Black-Scholes.xlsx


Values for the SAC call in the RIT H3 case.

9

4/16/2015

Hedging

The stock and the call move in the same direction.

They are very highly correlated.

Over short time intervals 𝜌 ≈ 1.

A short position in the call and a long position in the stock move in opposite directions.

They are very negatively correlated.

Over short time intervals 𝜌 ≈ −1.

Can we find a risk-free portfolio combination?


The H3 Case

We’re in a bank equity derivatives group.

A customer wants to buy a call option on SAC.

The customer would prefer to buy an exchange-traded call if one were available.

This isn’t possible, so the customer comes to us.

Acting as dealer, we sell to the customer.

In analyzing the trade, we concentrate on pricing and hedging.


10

4/16/2015

Pricing of the call

𝑋 = 50, 𝑟 = 0, 𝜎 = 0.15 (“15% per year”), 𝑇 = 20 𝑡𝑟𝑎𝑑𝑖𝑛𝑔 𝑑𝑎𝑦𝑠.

Annualize using trading days: 𝑇 =20

252= 0.079365

𝐶 = $0.843/𝑠ℎ𝑎𝑟𝑒

We need to factor in a profit for ourselves so we quote a price of $1.41 to the customer.

$1.41: “We price the call using 𝜎 = 0.25.”

The customer can do the calculations: they know the mark-up they’re paying.


Size

The customer wants to buy 200 calls.

The standard contract size is 100 shares.

All prices are quoted on a per-share basis, but when the customer buys, they pay us $1.41 × 100 × 200 = $28,200.


11

4/16/2015

Hedging

Immediately after the customer agrees to the trade, we are short 200 calls.

If the price of SAC falls, we’ll be okay. The call will expire unexercised.

If the price of SAC rises, the call will be exercised against us, costing us money.

We don’t want to make a directional bet on SAC. We want to hedge.


Can we hedge by buying 200 × 100 = 20,000 sh SAC?

“Worst case: SAC goes from $50 to, say, $100. At that point, the customer will exercise. They buy 20,000 from us at $50.”

“Shouldn’t we lock in a $50 purchase price for SAC by buying all the shares we might need right now?”


12

4/16/2015


46 48 50 52 54SAC Stock

1

2

3

4

5SAC Call Δ (“Delta”)=Slope=0.508

To hedge that we’ve written, we need to be long 0.508 shares of SAC

The recipe

Sell 200 100-share call options at $1.41

Receive cash of $28,200.

Buy 200 × 100 × 0.508 = 10,169 shares of stock

Pay with “borrowed” money

Are we really hedged?

What happens if 𝑆 changes by ± $0.01?


13

4/16/2015

The Black-Scholes calculations

Next: the mark-to-market position statements

27

28

S Assets Liabilities

50.00Cash received (20,000 ×$1.41)

28,200 16,857Call, mark-to-mkt (~20,000 ×

Stock (10,169𝑠ℎ × $50) 508,429 508,429 Loan / charge to capital

11,343 Net worth

49.99 Cash received 28,200 16,756Call, mark-to-mkt(~20,000 ×

Stock (10,169𝑠ℎ × $49.99) 508,327 508,429 Loan / charge to capital

11,343 Net worth

50.01 Cash received 28,200 16,959Call, mark-to-mkt(~20,000 ×

Stock (10,169𝑠ℎ × $50.01) 508,530 508,429 Loan / charge to capital

11,343 Net worth

14

4/16/2015

H3

To design the hedge, prepare a table that gives hedge ratios and the number of shares you should be long for SAC prices between $46 and $54 in $0.20 increments.

As the stock price changes, this is the target amount you should be trying to hold.

But adjusting the hedge too often, or using market orders will boost the trading costs.

Remember: the size of the position is 20,000 shares.


Practical complications

We’ve concentrated on hedging short term changes in the stock price S.

This is called delta hedging.

The hedge ratio also depends on 𝑇, 𝑟 and 𝜎.

𝜎 depends on market conditions and new information.

Even if 𝑟 and 𝜎 are constant, T will change with the passage of time.


15

4/16/2015

Adjusting the hedge requires us to trade in the direction of the market.

We sell when S falls, buy when S rises.

Do our trades push the price against us?

Are there other traders who are also trying to delta-hedge?

What if there is a significant news announcement when the market is closed?


Hedging and jumps

Delta hedging works best when successive price movements are small.

Slow accumulations of low-intensity information

Example SPY, April 15, 2011

Delta hedging does not work well when prices move due to large information shocks.


16

4/16/2015

Selling a call on ACOR, 14 April 2011


At 10AM, a customer wants a one-year ACOR call with 𝑋 = $20.

𝑇 = 1; assume 𝜎 = 0.5, 𝑟 = 0.

At 10AM, the stock price is 𝑆 ≈ $21.

From Black-Scholes 𝐶 = $2.976 𝑎𝑛𝑑 Δ = 𝑁 𝑑1 = 0.623.

We sell ten 100-share call options to a customer at $6

We hedge by buying 623 shares of ACOR.

At about 13:20 the stock price goes to $27.

At 𝑆 = $27, 𝐶 = $7.576 and Δ = 0.875


17

4/16/2015

Mark-to-market positions


Time Assets Liabilities

10:00 Cash receivedfrom sale of calls

6,000 2,976 Calls, mark-to-mkt $2.976 × 100 × 10

Stock (623 sh @ $21) 13,083 13,083 Loan / charge to capital

3,024 Net worth

14:00 Cash 6,000 7,576 Calls $7.576 × 100 × 10

Stock (623 sh @ $27) 16,821 13,083 Charge to capital

2,162 Net worth

Replication

An alternative direction for the H3 case.

The customer says, “The bank wants $1.41 for a call that Black-Scholes says is worth only $0.84? That’s crazy. Can we ‘manufacture’ the option ourselves?”

How does the customer do this?


trading to hedge: dynamic hedging - new york universitypeople.stern.nyu.edu/jhasbrou/teaching/post...

Documents